Information extraction apparatus, information extraction method, and information extraction program

An information extraction method comprising:

• acquiring a plurality of structured documents having a same specification and a different context;
• extracting, as variable elements, parts different between the acquired documents by calculating differences between a plurality of the structured documents;
• displaying a plurality of the extracted variable elements;
• receiving a selection as an extraction object which is at least one of the variable elements selected by a user via an input device extracted from each structured document;
• extracting, as peripheral information pieces, elements within a predetermined range from the respective variable elements extracted from each structured document;
• storing, in a storage unit, the variable elements and the peripheral information pieces with respect to at least the extraction object;
• re-acquiring a plurality of new structured documents, each new structured document having a second same specification and different context, where in the second same specification is different from the first same specification;
• re-extracting, as variable elements, parts different between the new structured documents;
• re-extracting, as peripheral information pieces, elements within the predetermined range from the respective re-extracted variable elements;
• calculating similarities of the variable elements and the peripheral information pieces between before and after the re-extraction, on the basis of (i) the re-extracted variable elements and the re-extracted peripheral information pieces and (ii) the variable elements and the peripheral information pieces stored in the storage unit, and
• identifying, out of the re-extracted variable elements, a variable element corresponding to the extraction object, on the basis of the calculated similarities.

A non-transitory computer readable medium including a program for causing a computer to execute the steps of:

• acquiring a plurality of structured documents having a same specification and a different context;
• extracting, as variable elements, parts different between the acquired documents by calculating differences between a plurality of the structured documents;
• displaying a plurality of the extracted variable elements;
• receiving a selection as an extraction object which is at least one of the variable elements selected by a user via an input device extracted from each structured document;
• extracting, as peripheral information pieces, elements within a predetermined range from the respective variable elements extracted from each structured document;
• storing, in a storage unit, the variable elements and the peripheral information pieces with respect to at least the extraction object;
• re-acquiring a plurality of new structured documents, each new structured document having a second same specification and different context, where in the second same specification is different from the first same specification;
• re-extracting, as variable elements, parts different between the new structured documents;
• re-extracting, as peripheral information pieces, elements within the predetermined range from the respective re-extracted variable elements;
• calculating similarities of the variable elements and the peripheral information pieces between before and after the re-extraction, on the basis of (i) the re-extracted variable elements and the re-extracted peripheral information pieces and (ii) the variable elements and the peripheral information pieces stored in the storage unit, and
• identifying, out of the re-extracted variable elements, a variable element corresponding to the extraction object, on the basis of the calculated similarities.